Bruce Tedesco, New modeling methods for simulating advertising impact
Bruce opened with a couple of observations:
1. Advertising analysis requires modeling methods based on dynamic principles
2. Consumer media consumption across channels leads to virtual days exceeding 24 hours
Dynamic models – agent-based models, which have been around for some time but only recently applied to marketing, simulate what happens among consumers in a virtual marketplace. Running not a single simulation, but many, we are able to gain empirical direction from studying the effects and interactions that arise when inputs, like advertising, are varied. Simulations are really about “emergent behavior.” If you’ve ever played the SIMS or a similar game, you’ve had experience with dynamic modeling.
“Rather than calculate,” Bruce said, “we compute. We don’t solve a single equation.” The emergent behavior arising from the simulations informs our analysis and strategy.
Discussion
How do you validate the simulations? There are ways to calibrate the simulations in real time, e.g. using multiple samples of the data. Best have been when we can run real time in market, and we can see that the results of the simulated marketplace match with the actual sales results in the market.
Theories have interesting validations – a lot of work in agent based modeling has been done with modeling the past. All of the conditions in Florence in 1400 were fed into a model, the same type of guilds appeared.
Real validations come in working with clients and sticking with it long enough.
Recent Comments